Controller Swap Drill

Question 1

Before de-terminating a failed controller, the first move is to photograph and label every wire and terminal — even when you're sure you'll remember the landings.

Accepted Answer

True. Always. A controller swap is mostly a re-landing job, and the field wiring is the part nobody documented well. Photos plus tape flags on each conductor turn "which wire went to UI-3?" from an afternoon of buzzing-out into a two-minute reference. Memory is the thing that fails an hour later when half the bundle is out of the panel and they all look the same.

Question 2

You've landed the field sensors (a 10K thermistor, an RTD, a 4-20 mA transmitter) on the new controller. What do they need before they'll read right?

Accepted Answer

Nothing on the sensor side — they're dumb. But the controller's input must be configured for the right type/range, and you verify wiring and loop power.. Field sensors carry no configuration — a thermistor is a resistor, a 4-20 mA transmitter just pushes current. The intelligence is in the controller's input: it has to know "UI-3 is a 10K type-2 thermistor" or "AI-5 is 4-20 mA, 0-250 °F" to turn the raw signal into a value. Land the wire, set the input type/range, confirm loop power and scaling — the sensor itself needs nothing.

Question 3

The new controller sits on an MS/TP (RS-485) trunk. Where do end-of-line termination resistors belong?

Accepted Answer

Only at the two physical ends of the trunk.. RS-485 wants exactly two terminations: one at each physical end of the daisy-chain, matched to the cable impedance to kill reflections. A device in the middle of the run is not terminated. If the controller you removed happened to be an end-of-line device, the new one needs its EOL jumper/resistor set the same way — and if you've mid-spanned a device that used to be the end, you've now got the wrong count and a flaky bus.

Question 4

Swapping an MS/TP controller, what should the new unit's MAC (station) address be set to?

Accepted Answer

The same station address the old unit used (via dip switch or config), so the supervisor and bus still find it where they expect.. Reuse the old station address. The supervisor's polling, the trunk's token-passing order, and any hard-coded references all expect the device at its known MAC. Pick a fresh address and you may collide with another master, fall outside Max_Master, or simply not appear where the integration looks for it. Set it on the new unit the same way the old one was set — dip switches or a config parameter.

Question 5

Valid MS/TP master station addresses run from 0 up to what maximum?

Accepted Answer

127. An MS/TP MAC is a single byte, and masters live in the range 0–127 (128–254 are reserved for slaves; 255 is the broadcast address). The practical ceiling on how high you should go is the bus's Max_Master setting — the token only passes up to Max_Master, so an address above it is invisible no matter how valid it looks.

Question 6

The replacement is a BACnet/IP controller. What should its device-instance number be?

Accepted Answer

The same device instance the old controller had — it's the network-wide identity the graphics and sequences reference.. The device instance is the controller's application-layer identity — what every graphic binding and inter-device reference points at, and what survives a re-IP or a move. Reuse the old instance and the rest of the system keeps working untouched. Pick a new one and you're re-binding everything that referenced the old device — and if the old instance is still claimed elsewhere, a duplicate-instance conflict is its own miserable afternoon.

Question 7

While you're working in the mechanical room on a controller swap, it's fine to back off a nearby balancing valve that's in your way and set it back "about where it was" after.

Accepted Answer

False. Leave balancing valves alone. A calibrated balancing valve's position is the result of a commissioning procedure, and "about where it was" is not a setting — disturb it and you've silently re-unbalanced that branch, with the complaint showing up weeks later in some occupied space. If a valve genuinely has to move, note its exact position first and restore it precisely, or flag it for re-balancing.

Question 8

The controller is failing but still communicates. What's the highest-value software move before you pull it?

Accepted Answer

Pull a full backup of its configuration and application program while it can still talk.. Get the backup first. A controller that still communicates is the easiest place to recover the as-built program, point list, tuning, schedules, and trims — all the site-specific work that isn't in any vendor default. Once the unit is dead or pulled, you're rebuilding from drawings and memory. A new controller ships blank; the value is in the configuration you're about to lose.

Question 9

After you download the application to the new controller, the BMS graphic for it still shows stale values or red ✗ marks. Likely reason?

Accepted Answer

The graphic's bindings point at device/point references that changed — they need re-pointing at the new controller's identity and point IDs.. Graphics bind to specific references — device instance plus object/point identifiers. If the swap changed the device instance, or the point IDs/names came out different after the download, every binding that resolved before now dangles. Reusing the old device instance avoids most of this; where IDs still moved, the bindings get re-pointed. Stale values and ✗ marks are the binding layer telling you it can't find what it used to.

Question 10

New controller in, program downloaded, and the graphic shows every point live with believable numbers. The tech calls the swap done. What did they likely skip? points on graphic: all reading, values look sane ✓ sequence verified: ? overrides / priority: ?

Accepted Answer

Verifying the sequence actually runs and re-checking commanded outputs: points reading ≠ logic working, and priority-array overrides / releases don't survive a swap.. Points reading values only proves the inputs are wired and scaled — it says nothing about whether the control logic does the right thing. You have to exercise the sequence: command outputs, watch the equipment respond, confirm interlocks and setpoints. And commandable outputs come up with their priority arrays reset — any operator override or hand-off that was holding a point before the swap is gone, and any slot you leave written will quietly fight the sequence. "It reads" is the start of commissioning, not the end.

Controller Swap Drill Field Drill